Model preporučiteljskog sustava temeljen na umjetnom imunološkom sustavu

scientific master's theses

• Title (english): Recommender system model based on artificial immune system
• Author: Branko Mihaljević
• Mentor: Mario Žagar (mentor)
• Committee member: Mario Žagar (član povjerenstva)
• Committee member: Mario Kovač (član povjerenstva)
• Committee member: Davor Antonić (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; (Department of Control and Computer Engineering); Zagreb
• Defense date and country: 2005-12-21, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing; Process Computing
• Universal decimal classification (UDC): 004 - Computer science and technology. Computing. Data processing

Abstract

Zbog svojih brojnih svojstava i principa funkcioniranja umjetni imunološki sustavi su primjenjivi u domeni problema klasifikacije i ocjenjivanja objekata s uočljivim značajkama. Detaljnim uvidom u mogućnosti primjene umjetnih imunoloških sustava zaključuje se da su pogodni za rješavanje problema preporuke zanimljivih sadržaja, poštujući evoluciju korisničkih sklonosti i interesa. Jedna od novih teorija kontekstno zavisnih odgovora u području imunoloških sustava, nazvana teorija opasnosti, objašnjava reakcije imunoloških stanica na patogene. Ovaj rad opisuje izgradnju modela preporučiteljskog sustava za potrebe preporuke sadržaja temeljenog na umjetnom imunološkom sustava uz korištenje teorije opasnosti. Analizom potreba preporučiteljskog sustava definirani su zahtjevi i svojstva modela, te je predložena arhitektura sustava. Na osnovu usporedbe s drugim umjetnim imunološkim sustavima, predloženo je detaljno radno okruženje kroz reprezentaciju komponenata sustava, odabir više različitih mjera sličnosti za usporedbu vektora značajki, i odabir procedura upravljanja dinamikom i životnim ciklusom komponenata sustava. Za potrebe modela, projektiran i izgrađen je algoritam RECAIS, te je mjerena njegova učinkovitost s realnim podacima. Model sustava je primijenjen na problem preporuke vijesti FERWeb portala te je diskutirana mogućnost primjene preporuke sadržaja u drugim sustavima za upravljanje sadržajem.

Abstract (english)

Because of their characteristics and principles of functioning, artificial immune systems are applicable in classification and rating problem domain on objects with perceptible features. Detail inquiry of AIS application possibilities concludes suitability for solving content recommendation problems, honoring evolving user preferences and interests. One of novel context dependant response theories in area of immune systems, called Danger theory, explains immunological cells reactions to pathogens. This thesis addresses recommender system model construction for content recommendation based on artificial immune system with Danger theory. Recommender system requirement analysis defines model requisites and properties, and system architecture is being proposed. Based on comparison with other artificial immune systems, detail model framework is proposed via system component representation, selection of different similarity measures for feature vector comparison, and selection of procedures governing system dynamics and components lifecycle. Regarding model requirements, algorithm RECAIS is designed and implemented, and its efficiency is measured with real data. System model is applied to news recommendation problem on FERWeb portal. Content recommendation application capabilities with other content management systems are discussed.

• Keywords: Preporučiteljski sustav; umjetni imunološki sustav; AIS; teorija opasnosti; preporuka
• Keywords (english): Recommender system; artificial immune system; AIS; danger theory; recommendation
• Language: croatian
• URN:NBN: urn:nbn:hr:168:001356
• Study programme: Title: Postgraduate master programe in computing; Study programme type: university; Study level: postgraduate; Academic / professional title: magistar inženjer računarstva (magistar inženjer računarstva)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2019-07-16 07:21:07